ErlasticSearch

An Erlang client for Elasticsearch.

Build and Run

Start a rebar3 shell

rebar3 shell

Create an index :

erlastic_search:create_index(<<"index_name">>).

{ok, [{<<"ok">>,true},{<<"acknowledged">>,true}]}

Index a document :

erlastic_search:index_doc(<<"index_name">>, <<"type">>, [{<<"key1">>, <<"value1">>}]).

{ok,[{<<"ok">>,true},
     {<<"_index">>,<<"index_name">>},
     {<<"_type">>,<<"type">>},
     {<<"_id">>,<<"T-EzM_yeTkOEHPL9cN5B2g">>},
     {<<"_version">>,1}]}

Index a document (providing a document id) :

erlastic_search:index_doc_with_id(<<"index_name">>, <<"type">>, <<"id1">>, [{<<"key1">>, <<"value1">>}]).

{ok,[{<<"ok">>,true},
     {<<"_index">>,<<"index_name">>},
     {<<"_type">>,<<"type">>},
     {<<"_id">>,<<"id1">>},
     {<<"_version">>,2}]}

Search for a document :

erlastic_search:search(<<"index_name">>, <<"type">>, <<"key1:value1">>).

{ok,[{<<"took">>,6},
     {<<"timed_out">>,false},
     {<<"_shards">>,
      [{<<"total">>,5},{<<"successful">>,5},{<<"failed">>,0}]},
     {<<"hits">>,
      [{<<"total">>,3},
       {<<"max_score">>,0.30685282},
       {<<"hits">>,
        [[{<<"_index">>,<<"index_name">>},
          {<<"_type">>,<<"type">>},
          {<<"_id">>,<<"T-EzM_yeTkOEHPL9cN5B2g">>},
          {<<"_score">>,0.30685282},
          {<<"_source">>,[{<<"key1">>,<<"value1">>}]}],
         [{<<"_index">>,<<"index_name">>},
          {<<"_type">>,<<"type">>},
          {<<"_id">>,<<"id1">>},
          {<<"_score">>,0.30685282},
          {<<"_source">>,[{<<"key1">>,<<"value1">>}]}],
         [{<<"_index">>,<<"index_name">>},
          {<<"_type">>,<<"type">>},
          {<<"_id">>,<<"MMNcfNHUQyeizDkniZD2bg">>},
          {<<"_score">>,0.30685282},
          {<<"_source">>,[{<<"key1">>,<<"value1">>}]}]]}]}]}

Testing

In another terminal use docker-compose to start an Elasticsearch instance :

docker-compose up

For convenience, you can also start a Kibana instance for analysis/visualization :

docker-compose -f docker-compose.yml -f docker-compose-kibana.yml up

Run Common Test:

rebar3 ct

Using another JSON library than jsx

By default, we assume all the JSON erlang objects passed to us are in jsx's representation. And similarly, all of Elasticsearch's replies will be decoded with jsx.

However, you might already be using another JSON library in your project, which might encode and decode JSONs from and to a different erlang representation. For example, jiffy:

1> SimpleJson = <<"{\"key\":\"value\"}">>.
<<"{\"key\":\"value\"}">>

2> jiffy:decode(SimpleJson).
{[{<<"key">>,<<"value">>}]}

3> jsx:decode(SimpleJson).
[{<<"key">>,<<"value">>}]

In that case, you probably want erlastic_search to use your JSON representation of choice instead of jsx's.

You can do so by defining the ERLASTIC_SEARCH_JSON_MODULE environment variable when compiling erlastic_search, for example:

export ERLASTIC_SEARCH_JSON_MODULE=jiffy
rebar compile

The only constraint is that ERLASTIC_SEARCH_JSON_MODULE should be the name of a module, in your path, that defines the two following callbacks:

-callback encode(erlastic_json()) -> binary().
-callback decode(binary()) -> erlastic_json().

where erlastic_json() is a type mapping to your JSON representation of choice.